A Systematic Review of Machine Learning in Credit Card Fraud Detection

Credit card fraud remains a major concern for global financial institutions, driving the need for effective detection systems. This paper presents a systematic review of 52 machine learning-based studies on credit card fraud detection published over the past decade (2013–2025), with a focus on the widely adopted MLG-ULB benchmark dataset. The review categorizes algorithms into traditional machine learning, deep learning, ensemble, and emerging methods, and compares them using standard performance metrics and deployment considerations. Ensemble and tree-based models consistently rank among the top-performing techniques, with Random Forest achieving up to 99.98% accuracy, while deep learning methods like Long Short-Term Memory networks excel at identifying temporal patterns but require significant computational resources. Emerging paradigms such as quantum machine learning and graph neural networks show promise but remain constrained by scalability and implementation complexity. The findings highlight a shift in research priorities toward improving model interpretability, real-time processing, and privacy-preserving learning to support practical deployment in the financial sector. Key limitations identified include dataset-specific constraints, class imbalance challenges, and the need for regulatory compliance in real-world implementations.